【摘要】 针对目前风量分布中依靠经验认识全部近似认为是正态分布的情况,利用概率图分析法给出了判断风量分布的严格定义,且给出了基于对数正态分布的风量分布模型,证明了并非所有风量分布均服从正态分布.为避开失效模式的NP难题,提出网络可靠度的近似算法.该方法以递推分解算法为基础,引入多维正态变量积分公式,计算多失效模型的联合概率,进而计算风网系统的可靠度,大大降低了计算复杂性.对于特别复杂的风网,也可以采用上下界逼近的方法得出近似值,并通过实例分析,验证了本方法的高精度和高效率.更多还原

【Abstract】 Current understanding of air flows assumes a normal distribution to the flows.A probabilistic graphical analysis is used to show the actual distribution to be more accurately given by a log normal distribution.Some air flows are quite decidedly not normally distributed.An approximation algorithm of air network reliability is proposed to avoid the NP failure mode.The method is based on recursive decomposition.A multidimensional,normal variable,integral is introduced for calculation of the joint probability of multiple failures.Reliability of the ventilation network is calculated and the complexity of the computations is greatly reduced.Complex networks may benefit from the method of approximation bounds.A practical example analysis is presented showing the high efficiency and accuracy of the method.更多还原